Running Title: Global change modulates coral physiology


Colleen B Bove1,2*, Sarah W Davies1, Justin B Ries3, James Umbanhowar2,4, Bailey C Thomasson4,5, Elizabeth B Farquhar2,6, Jess A McCoppin4, Karl D Castillo2,7

1 The Department of Biology, Boston University, Boston, Massachusetts, USA
2 Environment, Ecology, and Energy Program, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
3 Department of Marine and Environmental Sciences, Northeastern University, Nahant, MA, USA
4 The Department of Biology, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
5 Coral Restoration Foundation, Key Largo, Florida, USA
6 Center for Marine Science, University of North Carolina Wilmington, Wilmington, NC, USA
7 The Department of Marine Science, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA

*Corresponding author:


Abstract:

Global change driven by anthropogenic carbon emissions is altering ecosystems at unprecedented rates, especially coral reefs, whose symbiosis with algal symbionts is particularly vulnerable to increasing ocean temperatures and altered carbonate chemistry. Here, we assess the physiological responses of three Caribbean coral (animal host + algal symbiont) species from an inshore and offshore reef environment after exposure to simulated ocean warming (28, 31 °C), acidification (300 - 3290 µatm), and the combination of stressors for 93 days. We used multidimensional analyses to assess how a variety of coral physiological parameters respond to ocean acidification and warming. Our results demonstrate reductions in coral health in Siderastrea siderea and Porites astreoides in response to projected ocean acidification, while future warming elicited severe declines in Pseudodiploria strigosa. Offshore S. siderea fragments exhibited higher physiological plasticity than inshore counterparts, suggesting that this offshore population was more susceptible to changing conditions. There were no plasticity differences in P. strigosa and P. astreoides between natal reef environments, however, temperature evoked stronger responses in both species. Interestingly, while each species exhibited unique physiological responses to ocean acidification and warming, when data from all three species are modelled together, convergent stress responses to these conditions are observed, highlighting the overall sensitivities of tropical corals to these stressors. Our results demonstrate that while ocean warming is a severe acute stressor that will have dire consequences for coral reefs globally, chronic exposure to acidification may also impact coral physiology to a greater extent in some species than previously assumed. The species-specific responses to global change we observe will likely manifest in altered Caribbean reef assemblages in the future.

Keywords
coral physiology; ocean acidification; ocean warming; plasticity; symbiosis; global change


Citation:

Repository DOI hosted on Zenodo DOI





Principal component analyses

Methods:

Principal component analysis (PCA) (function prcomp) of scaled and centered physiological parameters (host carbohydrate, host lipid, host protein, algal symbiont chlorophyll a, algal symbiont cell density, calcification rate as previously for the same samples in Bove et al (2019)) were employed to assess the relationship between physiological parameters and treatment conditions for each coral species. Main effects (temperature, pCO2, reef environment) were evaluated with PERMANOVA using the adonis2 function (vegan package; version 2.5.7 (Oksanen et al., 2020)). No interactions between main effects were identified as significant, so interaction terms were dropped from each model resulting in fully additive models.


Siderastrea siderea

## Permutation test for adonis under reduced model
## Marginal effects of terms
## Permutation: free
## Number of permutations: 1500
## 
## adonis2(formula = sid_pca_df ~ fpco2 + ftemp + reef, data = s_df, permutations = bootnum, method = "eu", by = "margin")
##          Df SumOfSqs      R2      F    Pr(>F)    
## fpco2     3    59423 0.20282 7.9334 0.0006662 ***
## ftemp     1     9320 0.03181 3.7329 0.0546302 .  
## reef      1    24705 0.08432 9.8948 0.0013324 ** 
## Residual 80   199740 0.68174                     
## Total    85   292988 1.00000                     
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1



Pseudodiploria strigosa

## Permutation test for adonis under reduced model
## Marginal effects of terms
## Permutation: free
## Number of permutations: 1500
## 
## adonis2(formula = dip_pca_df ~ reef + ftemp + fpco2, data = p_df, permutations = bootnum, method = "eu", by = "margin")
##          Df SumOfSqs      R2       F    Pr(>F)    
## reef      1   101796 0.09008 14.8653 0.0006662 ***
## ftemp     1   519372 0.45958 75.8438 0.0006662 ***
## fpco2     3    30444 0.02694  1.4819 0.2225183    
## Residual 71   486202 0.43023                      
## Total    76  1130099 1.00000                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1



Porites astreoides

## Permutation test for adonis under reduced model
## Marginal effects of terms
## Permutation: free
## Number of permutations: 1500
## 
## adonis2(formula = por_pca_df ~ reef + ftemp + fpco2, data = a_df, permutations = bootnum, method = "eu", by = "margin")
##          Df SumOfSqs      R2       F    Pr(>F)    
## reef      1      724 0.00512  0.5264 0.4856762    
## ftemp     1    27051 0.19129 19.6601 0.0013324 ** 
## fpco2     3    30537 0.21594  7.3978 0.0006662 ***
## Residual 62    85309 0.60325                      
## Total    67   141417 1.00000                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1



Figure 1

Figure 1. Principal component analysis (PCA) of all coral physiological parameters for (A) S. siderea, (B) P. strigosa, and (C) P. astreoides after 93 days of exposure to different temperature and pCO2 treatments. PCAs in the top row are depicted by temperature treatment for each species (28\(^\circ\)C blue; 31\(^\circ\)C red) and the bottom row of PCAs are depicted by pCO2 for each species (pre industrial [300 \(\mu\)atm], light purple; current day [420 \(\mu\)atm], dark purple; end-of-century [680 \(\mu\)atm], light orange; extreme [3290 \(\mu\)atm], dark orange). Arrows represent significant (p < 0.05) correlation vectors for physiological parameters (rate = calcification rate; den = symbiont density; chla = chlorophyll a; pro = protein; carb = carbohydrate; lipid = lipid; sum/red = color intensity) and ellipses represent 95% confidence based on multivariate t-distributions.



Figure 4

## Permutation test for adonis under reduced model
## Terms added sequentially (first to last)
## Permutation: free
## Number of permutations: 1500
## 
## adonis2(formula = all_pca_df ~ fpco2 + ftemp + reef + species + ftemp:species + fpco2:species + reef:species, data = all_df, permutations = bootnum, method = "eu")
##                Df SumOfSqs      R2        F    Pr(>F)    
## fpco2           3   149393 0.03848   8.2405 0.0006662 ***
## ftemp           1    17313 0.00446   2.8650 0.0966023 .  
## reef            1    58058 0.01496   9.6075 0.0013324 ** 
## species         2  1642613 0.42313 135.9102 0.0006662 ***
## ftemp:species   2   553351 0.14254  45.7844 0.0006662 ***
## fpco2:species   6    90865 0.02341   2.5061 0.0206529 *  
## reef:species    2    77259 0.01990   6.3924 0.0026649 ** 
## Residual      214  1293204 0.33312                       
## Total         231  3882055 1.00000                       
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Figure 4. Principal component analysis (PCA) comparing the physiology of all three species at the end of the experiment depicted by (A) species, (B) pCO2 treatment, and (C) temperature treatment. Arrows represent significant (p < 0.05) correlation vectors for physiological parameters and ellipses represent 95% confidence based on multivariate t-distributions.



Results

Principal component analysis

Two principal components (PCs) explained approximately 66% of the variance in physiological responses of S. siderea to ocean acidification and warming treatments (Figure 1A). PC1 was driven by differences in algal symbiont physiology (chlorophyll a, cell density), while PC2 represented an inverse relationship between host energy reserves (lipid, protein, carbohydrate) and calcification rates and color intensities. Overall, higher pCO2 and temperature resulted in reduced S. siderea physiology (Figure 1A). Treatment pCO2 predominantly drove S. siderea physiological responses (p = 7e-04; Table S4), while temperature and reef environment did not explain as much variation in physiological responses (p = 0.05 and p = 0.001, respectively; Table S4; Figure S4A). These observed responses are driven by declines in total host physiology under warming as well as reduced symbiont physiology with increasing pCO2 (Figure S5A). Further, no significant interactive effect between temperature and pCO2 was detected in S. siderea physiology (Table S2; Figure 3D).

For P. strigosa, 74% of the variance in responses to treatments was explained by two PCs (Figure 1B). PC1 explained most of the variation of physiological parameters with the exception of host lipid content, which was represented in PC2. Physiology of P. strigosa was clearly reduced under warming (p = 7e-04; Table S4) and in offshore samples (p = 7e-04; Table S4; Figure S4B), however, pCO2 did not clearly alter physiology (Figure 1B; p = 0.2; Table S4). This clear decline in physiology under warming is driven by declines in symbiont physiology and total host protein content (Figure S5B). Again, no significant interactive effect between temperature and pCO2 was detected (Table S2; Figure 3E).

For P. astreoides, the first two PCs explained about 59% of the total variance in response to treatment (Figure 1C). Samples separated most clearly along PC1 driven primarily by calcification rate and algal symbiont density, while PC2 exhibited an inverse relationship between host total carbohydrate and color intensity. Overall, higher pCO2 reduced P. astreoides physiology, while elevated temperature resulted in improved physiology (Figure 1C). These patterns are most notable in the reduced host energy reserves in response to increasing pCO2 and higher symbiont physiology and lipid content under warming (Figure S5C). Temperature (p = 0.001; Table S4) and pCO2 (p = 7e-04; Table S4) clearly altered P. astreoides physiology, while reef environment was not significant (p = 0.5; Table S4; Figure S4C) and there was no significant interactive effect between temperature and pCO2 (Table S2; Figure 3F).


Species differences in coral physiology

The first two PCs of coral physiology explained about 62% of the total variance across samples (Figure 4). In general, fragments of S. siderea contained higher chlorophyll a content, host carbohydrate, and host lipid content, while P. strigosa fragments typically had greater host protein content accompanied by higher calcification rates, and fragments of P. astreoides were differentiated by their high symbiont densities (Figure 4A, S4; Table S7). Despite being different coral species, coral physiology exhibited similar declines in responses to increasing pCO2 treatments (Figure 4B; Table S7), however, responses to temperature were highly species-specific (Figure 4C; Table S7). Furthermore, corals from the inshore reef environment exhibited more constrained physiology than their offshore counterparts (Figure S20; Table S7).



Correlation assessments

Methods:

Correlations of all physiological parameters were assessed to determine the relationships between parameters within each species. The Pearson correlation coefficient (R2) of each comparison was calculated using the corrgram package (version 1.14 (Wright, 2018)) and the significance was calculated using the cor.test function. These relationships were then visualized through simple scatterplots.


Figure 2

Figure 2. Coral physiological parameter scatter plots (top) and correlation matrices (bottom) for (A) S. siderea, (B) P. strigosa, and (C) P. astreoides showing pairwise comparisons of within each species. Scatter plots of each pairwise combination of physiological parameters are displayed on the top with temperature treatment depicted by shape (28\(^\circ\)C closed points; 31\(^\circ\)C open points) and pCO2 treatment depicted by color (pre industrial [300 \(\mu\)atm], light purple; current day [420 \(\mu\)atm], dark purple; end-of-century [680 \(\mu\)atm], light orange; extreme [3290 \(\mu\)atm], dark orange). Strengths of the correlations (R2 via Pearson correlation coefficients) between each pairwise combination of physiological parameters are indicated by darker shades of blue on the bottom with significance depicted by asterisks according to significance level (* p < 0.05; ** p < 0.01; *** p < 0.001). R2 and significance levels correspond to the scatter plot at the intersection between two physiological parameters.



Results

Correlations of physiological parameters

Coral physiological parameters were generally positively correlated with one another within each of the three species. Correlations between S. siderea physiological parameters identified 15 significant relationships out of all 21 possible comparisons (Figure 2A). Of those significant correlations, six resulted in a Pearson’s correlation coefficient (R2) equal to or greater than 0.5, with the strongest relationship identified between symbiont density and chlorophyll a (R2 = 0.72).

All pairwise physiological parameters were significantly correlated with one another in P. strigosa and, of those, 15 correlations exhibit moderate (R2 > 0.50) positive relationships (Figure 2B). Notably, the two strongest correlations were host carbohydrate vs. host protein (R2 = 0.70) and host carbohydrate vs. chlorophyll a (R2 = 0.76).

Compared to both S. siderea and P. strigosa, fewer physiological traits were significantly (p < 0.05) correlated with one another in P. astreoides (12 significant out of 21 total comparisons; Figure 2C). Of the significant correlations, only two pairwise comparisons resulted in a Pearson’s correlation coefficient greater than 0.5: chlorophyll a vs. color intensity (R2 = 0.57) and host carbohydrate vs. host protein (R2 = 0.68).



Plasticity analyses

Methods:

Physiological plasticity of each experimental fragment was calculated for each species using all PCs calculated above as the distance between an experimental fragment and the control (420 μatm; 28°C) fragment from that same colony (Barott et al., 2021). The effects of treatment (pCO2 and temperature) and natal reef environment on calculated distances were assessed using generalized linear mixed effects models (function lmer) with a Gamma distribution and log-link and a random effect for colony (P. strigosa and P. astreoides) or tank crossed with colony (S. siderea). The best-fit model was selected as the model with the lowest AIC for each species (Table S2). Natal reef environment was only a significant predictor of plasticity in S. siderea; thus, samples were pooled across reef environments for both P. strigosa and P. astreoides. Parametric bootstraps were performed to model mean response and 95% confidence intervals with 1500 iterations and significant effects were defined as non-overlapping confidence intervals. Marginal and conditional R2 values of the best fit models were calculated using the r2_nakagawa function in the rcompanion package (version 2.4.1 (Mangiafico, 2021)). All figures and statistical analyses were carried out in R version 3.6.3 (R Core Team, 2018) and the accompanying data and code can be freely accessed on GitHub (github.com/seabove7/Bove_CoralPhysiology) and Zenodo (10.5281/zenodo.5093907).


Siderastrea siderea

## Generalized linear mixed model fit by maximum likelihood (Laplace
##   Approximation) [glmerMod]
##  Family: Gamma  ( log )
## Formula: dist ~ reef * fpco2 + ftemp + (1 | colony) + (1 | tank)
##    Data: sid_dist
## 
##      AIC      BIC   logLik deviance df.resid 
##    218.4    245.4    -97.2    194.4       58 
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.92132 -0.66729 -0.00254  0.42115  2.51068 
## 
## Random effects:
##  Groups   Name        Variance Std.Dev.
##  tank     (Intercept) 0.01357  0.1165  
##  colony   (Intercept) 0.03209  0.1791  
##  Residual             0.08867  0.2978  
## Number of obs: 70, groups:  tank, 21; colony, 11
## 
## Fixed effects:
##                  Estimate Std. Error  t value Pr(>|z|)    
## (Intercept)      1.050385   0.005617  187.007   <2e-16 ***
## reefN            0.009728   0.005650    1.722   0.0851 .  
## fpco2420         0.378300   0.005650   66.954   <2e-16 ***
## fpco2680         0.222648   0.005663   39.316   <2e-16 ***
## fpco23290        0.443555   0.005632   78.751   <2e-16 ***
## ftemp31          0.001769   0.005625    0.314   0.7531    
## reefN:fpco2420  -0.770729   0.005634 -136.809   <2e-16 ***
## reefN:fpco2680  -0.445519   0.005616  -79.326   <2e-16 ***
## reefN:fpco23290 -0.329693   0.005637  -58.483   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) reefN  fp2420 fp2680 f23290 ftmp31 rN:242 rN:268
## reefN       -0.003                                                 
## fpco2420     0.001 -0.002                                          
## fpco2680    -0.005  0.001 -0.001                                   
## fpco23290    0.003  0.001 -0.001  0.005                            
## ftemp31     -0.005 -0.002  0.000 -0.005  0.001                     
## rfN:fpc2420  0.001  0.001 -0.002  0.005 -0.001  0.001              
## rfN:fpc2680 -0.005 -0.002  0.002  0.003  0.005 -0.004  0.005       
## rfN:fp23290 -0.001 -0.002  0.002  0.001 -0.001  0.000  0.001  0.000
## optimizer (Nelder_Mead) convergence code: 0 (OK)
## Model failed to converge with max|grad| = 0.0126567 (tol = 0.002, component 1)
## # R2 for Mixed Models
## 
##   Conditional R2: 0.542
##      Marginal R2: 0.307



Pseudodiploria strigosa

## Generalized linear mixed model fit by maximum likelihood (Laplace
##   Approximation) [glmerMod]
##  Family: Gamma  ( log )
## Formula: dist ~ fpco2 + ftemp + (1 | colony)
##    Data: dip_dist
## 
##      AIC      BIC   logLik deviance df.resid 
##     97.5    104.8    -42.7     85.5       19 
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.70079 -0.72697  0.05779  0.77668  2.01138 
## 
## Random effects:
##  Groups   Name        Variance Std.Dev.
##  colony   (Intercept) 0.007898 0.08887 
##  Residual             0.137879 0.37132 
## Number of obs: 25, groups:  colony, 5
## 
## Fixed effects:
##             Estimate Std. Error t value Pr(>|z|)    
## (Intercept)  1.27912    0.14767   8.662   <2e-16 ***
## fpco2680    -0.33844    0.19338  -1.750   0.0801 .  
## fpco23290   -0.05892    0.18739  -0.314   0.7532    
## ftemp31      0.22671    0.17289   1.311   0.1898    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##           (Intr) fp2680 f23290
## fpco2680  -0.579              
## fpco23290 -0.578  0.484       
## ftemp31   -0.334  0.066  0.012
## # R2 for Mixed Models
## 
##   Conditional R2: 0.232
##      Marginal R2: 0.188



Porites astreoides

## Generalized linear mixed model fit by maximum likelihood (Laplace
##   Approximation) [glmerMod]
##  Family: Gamma  ( log )
## Formula: dist ~ fpco2 + ftemp + (1 | colony) + (1 | tank)
##    Data: por_dist
## 
##      AIC      BIC   logLik deviance df.resid 
##    142.4    158.3    -63.2    126.4       46 
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.6543 -0.5670 -0.1035  0.5470  2.2029 
## 
## Random effects:
##  Groups   Name        Variance Std.Dev.
##  tank     (Intercept) 0.001233 0.03512 
##  colony   (Intercept) 0.037406 0.19341 
##  Residual             0.056156 0.23697 
## Number of obs: 54, groups:  tank, 21; colony, 11
## 
## Fixed effects:
##             Estimate Std. Error t value Pr(>|z|)    
## (Intercept)  1.03344    0.12525   8.251  < 2e-16 ***
## fpco2420    -0.04516    0.11478  -0.393 0.693975    
## fpco2680     0.03277    0.07852   0.417 0.676454    
## fpco23290    0.12410    0.08225   1.509 0.131342    
## ftemp31      0.26204    0.06856   3.822 0.000132 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##           (Intr) fp2420 fp2680 f23290
## fpco2420  -0.153                     
## fpco2680  -0.318  0.380              
## fpco23290 -0.353  0.306  0.526       
## ftemp31   -0.166 -0.379 -0.073  0.068
## # R2 for Mixed Models
## 
##   Conditional R2: 0.493
##      Marginal R2: 0.145



Figure 3

Figure 3. Coral physiological plasticity of (A) S. siderea, (B) P. strigosa, and (C) P. astreoides after 93-day exposure to experimental treatments. Higher values represent greater plasticity in coral samples. Natal reef environment is depicted along the x axis for S. siderea, however, P. strigosa and and P. astreoides samples were pooled by reef environment. pCO2 treatment is depicted by color and shape (pre industrial [300 \(\mu\)atm], light purple; current day [420 \(\mu\)atm], dark purple; end-of-century [680 \(\mu\)atm], light orange; extreme [3290 \(\mu\)atm], dark orange) and temperature is represented as either closed (28 \(^\circ\)C) or open (31 \(^\circ\)C) symbols. The current day at 28 °C treatment is not depicted here since plasticity is represented as the distance from this treatment (420 μatm at 28 °C). Symbols and bars indicate modeled means and 95% confidence intervals.



Results

Coral physiological plasticity

Physiological plasticity of offshore S. siderea fragments exhibited a positive linear trend with increasing pCO2, while the inshore fragments appear to respond in a parabolic pattern to pCO2, with the lowest calculated distances occurring at 420 \(\mu\)atm, 31\(^\circ\)C and 680 \(\mu\)atm, 28\(^\circ\)C (Figure 3A). Further, offshore S. siderea fragments exhibited higher plasticity in the extreme pCO2 treatment than in inshore fragments reared in the pre-industrial, current-day, and extreme pCO2 treatments, regardless of temperature (Figure 3A; Table S5).

Plasticity of P. strigosa and P. astreoides was not clearly different between colonies based on natal reef environments (see Table S3). No clear differences in physiological plasticity in response to treatment were identified in P.strigosa (Figure 3B; Table S5), however, this is likely due to reduced sample sizes in this analysis as a result of only five colonies (Noffshore = 3, Ninshore = 2) present in the control treatment for distance calculations.

Elevated temperatures generally resulted in higher plasticity of P. astreoides compared to control temperatures (Figure 3C; Table S5), however, this trend was not clearly different within each pCO2 treatment. Physiological plasticity of P. astreoides was significantly lower in both the pre-industrial and end-of-century pCO2 treatments at control temperatures than that measured in the extreme pCO2 treatment combined with the elevated temperature.



Supplemental Figures

Figure S1

## [1] "vobjtovarid4: **** WARNING **** I was asked to get a varid for dimension named record BUT this dimension HAS NO DIMVAR! Code will probably fail at this point"

Figure S1. Monthly MODIS satellite SST data from 2002 to 2021 for both the inshore (Port Honduras Marine Reserve; yellow) and the offshore (Sapodilla Cayes Marine Reserve; green) coral collection locations. Solid horizontal lines represent corresponding reef environment mean SST across duration. The blue dashed line represents the experimental control treatment temperature (28 \(^\circ\)C) and the red dashed line represents the experimental elevated temperature treatment (31 \(^\circ\)C). Note the temperature variability of the inshore site exceeding the offshore location. [Data accessability: NASA OBPG. 2020. MODIS Aqua Global Level 3 Mapped SST. Ver. 2019.0. PO.DAAC, CA, USA. Dataset accessed [2021-02-02] at https://doi.org/10.5067/MODSA-MO4D9]



Figure S2

Figure S2. Diagram showing allocation of coral fragments for a single species throughout the experiment. Color represents a different colony and shape represents reef environment. Four colonies (two from each reef environment) are reared within each tank (grey box), with three tanks comprising a treatment (white box). This is repeated for each pCO2 treatment at both temperatures. This same experimental design was used for all species. Figure taken from Bove et al. 2019.



Figure S3

Figure S3. Calculated and measured seawater parameters over the entire experimental period.



Figure S4

Figure S4. Principal component analysis (PCA) of all coral physiological parameters for S. siderea, P. strigosa, and P. astreoides depicted by natal reef environment (A-C; offshore green, inshore yellow) and the combination of pCO2 and temperature treatment (D-F). Arrows represent significant (p < 0.05) correlation vectors for physiological parameters and ellipses represent 95% confidence based on multivariate t-distributions.



Figure S5

Figure S5. Mean (\(\pm\)SE) physiological parameter (each row) measured for (A) S. siderea, (B) P. strigosa, and (C) P. astreoides at the completion of the 93-day experimental period. pCO2 treatment is represented along the x axis and the temperature is depicted by color (28\(^\circ\)C blue; 31\(^\circ\)C red).



Figure S6

Figure S6. Principal component analysis (PCA) of S. siderea coral host (protein, lipid, carbohydrate; left) or algal symbiont (chlorophyll a, symbiont density, color intensity; right) physiological parameters by temperature (28\(^\circ\)C blue; 31\(^\circ\)C red), pCO2 (pre industrial [300 \(\mu\)atm], light purple; current day [420 \(\mu\)atm], dark purple; end-of-century [680 \(\mu\)atm], light orange; extreme [3290 \(\mu\)atm], dark orange), and natal reef environment (offshore green; inshore yellow). Arrows represent significant (p < 0.05) correlation vectors for physiological parameters and ellipses represent 95% confidence based on multivariate t-distributions.



Figure S7

Figure S7. Principal component analysis (PCA) of P. strigosa coral host (protein, lipid, carbohydrate; left) or algal symbiont (chlorophyll a, symbiont density, color intensity; right) physiological parameters by temperature (28\(^\circ\)C blue; 31\(^\circ\)C red), pCO2 (pre industrial [300 \(\mu\)atm], light purple; current day [420 \(\mu\)atm], dark purple; end-of-century [680 \(\mu\)atm], light orange; extreme [3290 \(\mu\)atm], dark orange), and natal reef environment (offshore green; inshore yellow). Arrows represent significant (p < 0.05) correlation vectors for physiological parameters and ellipses represent 95% confidence based on multivariate t-distributions.



Figure S8

Figure S8. Principal component analysis (PCA) of P. asteroides coral host (protein, lipid, carbohydrate; left) or algal symbiont (chlorophyll a, symbiont density, color intensity; right) physiological parameters by temperature (28\(^\circ\)C blue; 31\(^\circ\)C red), pCO2 (pre industrial [300 \(\mu\)atm], light purple; current day [420 \(\mu\)atm], dark purple; end-of-century [680 \(\mu\)atm], light orange; extreme [3290 \(\mu\)atm], dark orange), and natal reef environment (offshore green; inshore yellow). Arrows represent significant (p < 0.05) correlation vectors for physiological parameters and ellipses represent 95% confidence based on multivariate t-distributions.



Figure S9

Figure S9. Coral host vs. algal symbiont physiological plasticity of (A) S. siderea, (B) P. strigosa, and (C) P. astreoides after 93-day exposure to experimental treatments. Higher values represent greater plasticity in coral samples. pCO2 treatment is depicted by color and shape (pre industrial [300 μatm], light purple; current day [420 μatm], dark purple; end-of-century [680 μatm], light orange; extreme [3290 μatm], dark orange) and temperature is represented as either closed (28 °C) or open (31 °C) symbols. Symbols and bars indicate modeled means and 95% confidence intervals.



Figure S10

Figure S10. Coral color changes over the experimental period. Representative images of fragments of (A) P. astreoides, (B) S. siderea, and (C) P. strigosa from the same colonies demonstrating change in coral color over time in either control (420 μatm; 28 °C) or warming (420 μatm; 31 °C) treatments from the start of the experiment (T0) to the end (T90).



Figure S11

Figure S11. Principal component analysis (PCA) comparing the physiology of all three species at the end of the experiment depicted by (A) reef environment and (B) combined pCO2 and temperature treatment. Arrows represent significant (p < 0.05) correlation vectors for physiological parameters and ellipses represent 95% confidence based on multivariate t-distributions.



Supplemental Tables

Table S1

Table S1. Number samples per species per treatment assessed in physiological and plasticity analyses. Note that plasticity sample sizes are smaller due to comparison within colony resulting in reduced samples when a control fragment was not present.

Physiology N
Plasticity N
Treatment Offshore Inshore Offshore Inshore
Porites astreoides
current day (uatm) 28C 6 6
current day (uatm) 31C 2 4 2 4
end-of-century (uatm) 28C 6 6 5 6
end-of-century (uatm) 31C 0 4 0 4
extreme (uatm) 28C 5 5 5 5
extreme (uatm) 31C 3 5 3 5
pre industrial (uatm) 28C 6 5 5 5
pre industrial (uatm) 31C 3 3 2 3
Pseudodiploria strigosa
current day (uatm) 28C 3 2
current day (uatm) 31C 3 2 0 0
end-of-century (uatm) 28C 9 6 4 2
end-of-century (uatm) 31C 5 3 2 1
extreme (uatm) 28C 8 6 3 2
extreme (uatm) 31C 3 2 2 0
pre industrial (uatm) 28C 10 6 4 2
pre industrial (uatm) 31C 5 4 2 1
Siderastrea Siderea
current day (uatm) 28C 6 5
current day (uatm) 31C 5 6 5 5
end-of-century (uatm) 28C 6 6 6 5
end-of-century (uatm) 31C 6 6 6 5
extreme (uatm) 28C 7 5 7 4
extreme (uatm) 31C 6 5 6 4
pre industrial (uatm) 28C 6 4 6 4
pre industrial (uatm) 31C 3 4 3 4



Table S2

Table S2. PERMANOVA model assessment for best-fit model selection of PCAs per species. Akaike information criterion (AIC) was used to select the best-fit model per species. For all species, the fully additive model was the best-fit model (temperature + pCO2 + reef).

Species Full interactive model AIC Best fit (additive) model AIC
S. siderea 692.3 678.5
P. strigosa 696.9 685.8
P. astreoides 500.5 497.1



Table S3

Table S3. Model performance comparisons of generalized linear mixed effects models (GLMM) for plasticity assessments to select the best-fit model per species using the package performance (version 0.7.3). Akaike information criterion (AIC) was used to select the best-fit model per species. The models highlighted in grey were used for bootstrapping estimates and 95% confidence intervals.

Model formula AIC Conditional R2 Marginal R2
Siderastrea Siderea
reef environment * pCO2 * temperature + (1 | colony) 223.2 0.545 0.366
reef environment * pCO2 + temperature + (1 | colony) 218.8 0.506 0.322
reef environment * pCO2 + temperature + (1 | colony) + (1 | tank) 218.4 0.542 0.307
reef environment + pCO2 * temperature + (1 | colony) 225.6 0.442 0.253
reef environment + pCO2 + temperature + (1 | colony) 221.6 0.442 0.254
reef environment * (pCO2 + temperature) + (1 | colony) 220.1 0.511 0.329
pCO2 + temperature + (1 | colony) 222.1 0.370 0.088
Pseudodiploria strigosa
reef environment * pCO2 * temperature + (1 | colony) 110.7 0.427 0.347
reef environment * pCO2 + temperature + (1 | colony) 106.0 0.341 0.292
reef environment + pCO2 * temperature + (1 | colony) 106.9 0.313 0.271
pCO2 + temperature + (1 | colony) 102.8 0.268 0.224
Porites astreoides
reef environment * pCO2 * temperature + (1 | colony) 153.1 0.527 0.199
reef environment * pCO2 + temperature + (1 | colony) 145.9 0.521 0.195
reef environment + pCO2 * temperature + (1 | colony) 146.2 0.500 0.174
reef environment + pCO2 + temperature + (1 | colony) 142.3 0.499 0.174
reef environment * (pCO2 + temperature) + (1 | colony) 147.9 0.522 0.195
pCO2 + temperature + (1 | colony) 140.4 0.485 0.147
pCO2 + temperature + (1 | colony) + (1 | tank) 142.4 0.493 0.145





Table S4

Table S4. PERMANOVA model output from each species using the adonis2 function with 1500 iterations.
Df Sum of Squares R2 F P-value
Siderastrea Siderea
pCO2 3 59423 0.203 7.93 0.00067
temperature 1 9320 0.032 3.73 0.05463
reef environment 1 24705 0.084 9.89 0.00133
Residual 80 199740 0.682
Total 85 292988 1.000
Pseudodiploria strigosa
reef environment 1 101796 0.090 14.87 0.00067
temperature 1 519372 0.460 75.84 0.00067
pCO2 3 30444 0.027 1.48 0.22252
Residual 71 486202 0.430
Total 76 1130099 1.000
Porites astreoides
reef environment 1 724 0.005 0.53 0.48568
temperature 1 27051 0.191 19.66 0.00133
pCO2 3 30537 0.216 7.40 0.00067
Residual 62 85309 0.603
Total 67 141417 1.000





Table S5

Table S5. GLMM output from plasticity assessments for each species. The intercept of each model was set as 300 \(\mu\)atm, 28\(^\circ\)C, and inshore reef environment.
Estimate Standard error Statistic P-value
Siderastrea Siderea
(Intercept) 1.050 0.006 187.01 0.000
reef environment (offshore) 0.010 0.006 1.72 0.085
pCO2-current 0.378 0.006 66.95 0.000
pCO2-EOC 0.223 0.006 39.32 0.000
pCO2-extreme 0.444 0.006 78.75 0.000
temperature (31C) 0.002 0.006 0.31 0.753
reef environment (offshore):pCO2-current -0.771 0.006 -136.81 0.000
reef environment (offshore):pCO2-EOC -0.446 0.006 -79.33 0.000
reef environment (offshore):pCO2-extreme -0.330 0.006 -58.48 0.000
Conditional R2 0.542
Marginal R2 0.307
Pseudodiploria strigosa
(Intercept) 1.279 0.148 8.66 0.000
pCO2-EOC -0.338 0.193 -1.75 0.080
pCO2-extreme -0.059 0.187 -0.31 0.753
temperature (31C) 0.227 0.173 1.31 0.190
Conditional R2 0.232
Marginal R2 0.188
Porites astreoides
(Intercept) 1.033 0.125 8.25 0.000
pCO2-current -0.045 0.115 -0.39 0.694
pCO2-EOC 0.033 0.079 0.42 0.676
pCO2-extreme 0.124 0.082 1.51 0.131
temperature (31C) 0.262 0.069 3.82 0.000
Conditional R2 0.493
Marginal R2 0.145





Table S6

Table S6. PERMANOVA model output across species using the adonis2 function with 1500 iterations.
Df Sum of Squares R2 F P-value
pCO2 3 149393 0.038 8.24 0.00067
temperature 1 17313 0.004 2.87 0.09660
reef environment 1 58058 0.015 9.61 0.00133
species 2 1642613 0.423 135.91 0.00067
temperature:species 2 553351 0.143 45.78 0.00067
pCO2:species 6 90865 0.023 2.51 0.02065
reef environment:species 2 77259 0.020 6.39 0.00266
Residual 214 1293204 0.333
Total 231 3882055 1.000





Table S7

Table S7. PERMANOVA model output of coral host or algal symbiont physiology per species using the adonis2 function with 1500 iterations depicted in Figures S2-S4.
Coral host
Algal symbionts
Df Sum of Squares R2 F P-value Df Sum of Squares R2 F P-value
Siderastrea Siderea
pCO2 3 1 0.019 0.55 0.77615 3 61056 0.208 8.15 0.00067
temperature 1 3 0.075 6.65 0.00200 1 7468 0.025 2.99 0.09061
reef environment 1 0 0.006 0.52 0.58894 1 24705 0.084 9.90 0.00266
Residual 80 30 0.901 80 199684 0.682
Total 85 34 1.000 85 292913 1.000
Pseudodiploria strigosa
pCO2 3 1 0.041 1.23 0.26382 3 26899 0.024 1.31 0.27848
temperature 1 3 0.147 13.12 0.00067 1 515173 0.456 75.24 0.00067
reef environment 1 0 0.020 1.75 0.17255 1 101793 0.090 14.87 0.00200
Residual 71 14 0.793 71 486140 0.430
Total 76 18 1.000 76 1130005 1.000
Porites astreoides
pCO2 3 2 0.136 3.48 0.01666 3 29037 0.205 7.04 0.00067
temperature 1 0 0.036 2.76 0.10193 1 26338 0.186 19.15 0.00067
reef environment 1 0 0.021 1.64 0.19920 1 724 0.005 0.53 0.48168
Residual 62 10 0.807 62 85288 0.603
Total 67 13 1.000 67 141387 1.000




Table S8

Table S8. Model performance comparisons of generalized linear mixed effects models (GLMM) for plasticity assessments to select the best-fit model comparing host and symbiont using the package performance (version 0.7.3). Akaike information criterion (AIC) was used to select the best-fit model. The models highlighted in grey were used for bootstrapping estimates and 95% confidence intervals.
Model formula AIC Conditional R2 Marginal R2
species * part * reef environment * pCO2 * temperature + (1 | colony) 923.1 0.430 0.318
species * part * reef environment * pCO2 + temperature + (1 | colony) 904.6 0.338 0.225
species * part * reef environment + pCO2 * temperature + (1 | colony) 890.2 0.222 0.113
species * reef environment + part + pCO2 + temperature + (1 | colony) 881.2 0.205 0.095
species * part * reef environment * (pCO2 + temperature) + (1 | colony) 902.1 0.391 0.279
species * part * pCO2 + temperature + (1 | colony) 890.2 0.235 0.093
reef environment * species * part * (pCO2 + temperature) + (1 | colony) 902.1 0.391 0.279
reef environment * species * part * (pCO2 + temperature) + (1 | colony) 902.1 0.391 0.279
species * part * (pCO2 + temperature) + (1 | colony) 882.0 0.280 0.139
species * part * (pCO2 + temperature) + reef environment + (1 | colony) 883.5 0.282 0.144
species * part * (pCO2 + temperature + reef environment) + (1 | colony) 883.9 0.323 0.212





Table S9

Table S9. GLMM output from plasticity assessments for coral host vs. algal symbiont for each species. The intercept of each model was set as coral host, S. siderea, 300 \(\mu\)atm, and 28\(^\circ\)C.
Estimate Standard error Statistic P-value
(Intercept) 0.800 0.172 4.66 0.000
PSTR 0.199 0.290 0.69 0.492
PAST -0.357 0.244 -1.46 0.143
symbionts -0.515 0.185 -2.78 0.005
pCO2-current 0.234 0.214 1.10 0.273
pCO2-EOC 0.005 0.164 0.03 0.974
pCO2-extreme -0.021 0.161 -0.13 0.895
temperature (31C) -0.252 0.128 -1.96 0.050
PSTR:symbionts 0.223 0.308 0.72 0.470
PAST:symbionts 0.685 0.270 2.53 0.011
PSTR:pCO2-current -0.088 0.592 -0.15 0.882
PAST:pCO2-current -0.568 0.339 -1.68 0.094
PSTR:pCO2-EOC -0.450 0.298 -1.51 0.131
PAST:pCO2-EOC 0.051 0.240 0.21 0.833
PSTR:pCO2-extreme -0.154 0.288 -0.53 0.593
PAST:pCO2-extreme 0.236 0.245 0.96 0.335
PSTR:temperature (31C) 0.072 0.250 0.29 0.774
PAST:temperature (31C) 0.608 0.200 3.04 0.002
symbionts:pCO2-current -0.299 0.298 -1.00 0.316
symbionts:pCO2-EOC 0.134 0.227 0.59 0.555
symbionts:pCO2-extreme 0.564 0.226 2.50 0.013
symbionts:temperature (31C) 0.524 0.181 2.89 0.004
PSTR:symbionts:pCO2-current 0.362 0.818 0.44 0.659
PAST:symbionts:pCO2-current 0.673 0.473 1.42 0.154
PSTR:symbionts:pCO2-EOC -0.179 0.418 -0.43 0.669
PAST:symbionts:pCO2-EOC -0.180 0.335 -0.54 0.590
PSTR:symbionts:pCO2-extreme -0.482 0.401 -1.20 0.229
PAST:symbionts:pCO2-extreme -0.890 0.341 -2.61 0.009
PSTR:symbionts:temperature (31C) 0.189 0.350 0.54 0.589
PAST:symbionts:temperature (31C) -0.607 0.281 -2.16 0.031
Conditional R2 0.280
Marginal R2 0.139





Session information

Session information from the last run date on 2022-03-01:

## R version 3.6.3 (2020-02-29)
## Platform: x86_64-apple-darwin15.6.0 (64-bit)
## Running under: macOS Catalina 10.15.7
## 
## Matrix products: default
## BLAS:   /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRblas.0.dylib
## LAPACK: /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRlapack.dylib
## 
## locale:
## [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
## 
## attached base packages:
## [1] grid      stats     graphics  grDevices utils     datasets  methods  
## [8] base     
## 
## other attached packages:
##  [1] ggrepel_0.9.1      xts_0.12.1         zoo_1.8-9          raster_3.4-13     
##  [5] sp_1.4-5           ncdf4_1.17         janitor_2.1.0      rcompanion_2.4.1  
##  [9] car_3.0-11         carData_3.0-4      png_0.1-7          MASS_7.3-54       
## [13] performance_0.7.3  wesanderson_0.3.6  RColorBrewer_1.1-2 gridGraphics_0.5-1
## [17] corrplot_0.90      Hmisc_4.5-0        Formula_1.2-4      survival_3.2-12   
## [21] magick_2.5.2       ggpubr_0.4.0       vroom_1.5.4        lmerTest_3.1-3    
## [25] lme4_1.1-27.1      Matrix_1.3-4       kableExtra_1.3.4   finalfit_1.0.3    
## [29] ggfortify_0.4.14   cowplot_1.1.1      Rmisc_1.5          shiny_1.7.1       
## [33] vegan_2.5-7        lattice_0.20-44    permute_0.9-5      forcats_0.5.1     
## [37] stringr_1.4.0      purrr_0.3.4        tibble_3.1.3       tidyverse_1.3.1   
## [41] plotly_4.9.4.1     openxlsx_4.2.4     corrgram_1.14      tidyr_1.1.3       
## [45] ggbiplot_0.55      scales_1.1.1       plyr_1.8.6         dplyr_1.0.7       
## [49] ggplot2_3.3.5      broom_0.7.9        readr_2.0.1        knitr_1.33        
## 
## loaded via a namespace (and not attached):
##   [1] utf8_1.2.2          tidyselect_1.1.1    htmlwidgets_1.5.3  
##   [4] munsell_0.5.0       codetools_0.2-18    withr_2.4.2        
##   [7] colorspace_2.0-2    highr_0.9           rstudioapi_0.13    
##  [10] stats4_3.6.3        DescTools_0.99.42   ggsignif_0.6.2     
##  [13] labeling_0.4.2      farver_2.1.0        bit64_4.0.5        
##  [16] vctrs_0.3.8         generics_0.1.0      TH.data_1.0-10     
##  [19] xfun_0.25           R6_2.5.1            assertthat_0.2.1   
##  [22] promises_1.2.0.1    multcomp_1.4-17     nnet_7.3-16        
##  [25] rootSolve_1.8.2.2   gtable_0.3.0        multcompView_0.1-8 
##  [28] lmom_2.8            sandwich_3.0-1      rlang_0.4.11       
##  [31] systemfonts_1.0.2   splines_3.6.3       rgdal_1.5-23       
##  [34] rstatix_0.7.0       lazyeval_0.2.2      checkmate_2.0.0    
##  [37] yaml_2.2.1          abind_1.4-5         modelr_0.1.8       
##  [40] backports_1.2.1     httpuv_1.6.2        tools_3.6.3        
##  [43] ellipsis_0.3.2      jquerylib_0.1.4     proxy_0.4-26       
##  [46] Rcpp_1.0.7          base64enc_0.1-3     rpart_4.1-15       
##  [49] haven_2.4.3         cluster_2.1.2       fs_1.5.0           
##  [52] magrittr_2.0.1      data.table_1.14.0   lmtest_0.9-38      
##  [55] reprex_2.0.1        mvtnorm_1.1-2       matrixStats_0.57.0 
##  [58] hms_1.1.0           mime_0.11           evaluate_0.14      
##  [61] xtable_1.8-4        rio_0.5.27          jpeg_0.1-9         
##  [64] readxl_1.3.1        gridExtra_2.3       compiler_3.6.3     
##  [67] mice_3.13.0         crayon_1.4.1        minqa_1.2.4        
##  [70] htmltools_0.5.2     mgcv_1.8-36         later_1.3.0        
##  [73] tzdb_0.1.2          libcoin_1.0-8       expm_0.999-6       
##  [76] Exact_2.1           lubridate_1.7.10    DBI_1.1.1          
##  [79] dbplyr_2.1.1        boot_1.3-28         cli_3.0.1          
##  [82] parallel_3.6.3      insight_0.14.3      pkgconfig_2.0.3    
##  [85] numDeriv_2016.8-1.1 coin_1.4-1          foreign_0.8-75     
##  [88] xml2_1.3.2          svglite_2.0.0       bslib_0.3.1        
##  [91] webshot_0.5.2       rvest_1.0.1         snakecase_0.11.0   
##  [94] digest_0.6.27       rmarkdown_2.10      cellranger_1.1.0   
##  [97] htmlTable_2.2.1     nortest_1.0-4       gld_2.6.2          
## [100] curl_4.3.2          modeltools_0.2-23   nloptr_1.2.2.2     
## [103] lifecycle_1.0.0     nlme_3.1-151        jsonlite_1.7.2     
## [106] viridisLite_0.4.0   fansi_0.5.0         pillar_1.6.2       
## [109] fastmap_1.1.0       httr_1.4.2          glue_1.4.2         
## [112] zip_2.2.0           bit_4.0.4           class_7.3-19       
## [115] stringi_1.7.3       sass_0.4.0          latticeExtra_0.6-29
## [118] e1071_1.7-8